Universal Access To All Knowledge
Home Donate | Store | Blog | FAQ | Jobs | Volunteer Positions | Contact | Bios | Forums | Projects | Terms, Privacy, & Copyright
Search: Advanced Search
Anonymous User (login or join us)
Upload

Reply to this post | See parent post | Go Back
View Post [edit]

Poster: Administrator, Curator, or StaffBrad Leblanc Date: Jun 16, 2005 11:44pm
Forum: etree Subject: Re: How to automate FLAC file tagging

I good idea. I too would love to see this implemented.

the potential drawback of adding tags to FLAC files is that changes to the tags will alter the file's md5 checksum

Well, this won't be an issue on the way into the collection as long as we do:

1. Verify provided MD5.
2. Once it passes, tag the FLAC's with files.xml info.
3. Calculate new FLAC MD5.
4. Overwrite old MD5 with new one.

I'll add this to our feature request list.

I'd love to see the majority of seeders add tags to new FLAC seeds. However, in order for this to happen, it needs to be a simple process that doesn't require much more effort than creating the original text file.

I have a beta copy of software called "EasyFlac" that makes this *really* easy when creating FLAC's. It's basically a utility very similar to MKW but it's for FLAC - where it incorporates a ton of right clicking options. It has a built in tool to tag files similar to FLAC Frontend - but it has the additional option to automtically tag track names too based on a text file. I used it last night with this seed and it worked like a charm. A few of us have asked the author to add SHN support too. :)

I'll ping him in the hopes he can comment on plans for the software and whether it will be publicly released.

Reply to this post
Reply [edit]

Poster: Administrator, Curator, or StaffDiana Hamilton Date: Jun 17, 2005 12:18am
Forum: etree Subject: Re: How to automate FLAC file tagging

Interesting idea! You'd probably have to have an extra flag on the import form for "These are flac files that are already tagged, don't autochange the tags" or something like that.

Reply to this post
Reply [edit]

Poster: Administrator, Curator, or StaffMatthew Vernon Date: Jun 17, 2005 3:35am
Forum: etree Subject: Re: How to automate FLAC file tagging

There are two places where the md5's exist. Here and on db.etree.org.

Charles suggested we try to do the tagging when we converted the Phil Lesh collection to flac for the nonlossy format.

The issue I had in doing this went something as follows:

Step 1 - Confirm that the seed matched the md5 in db.etree.org for the source its purported to be in the info file

Step 2 - Convert that to flac using the scripts we cobbled together, which verified that the audio portion was digitally identical for both the shn and flac versions

step 3 - Manually add the flac-md5 signatures to the corresponding etree shnid entry so that etree will link to the source on the IA and the metadata entry scripts will preload setlist venue and other fields using the information in db.etree.org

Step 4 - Enter the metadata on IA

Normally at this point we would submit the seed to the Contribution Engine and that would be it.

To add the metadata tags to the flac source and not lose ANY linking between etree<->IA it sould seem that the following would have to be done.

Step 5 - After the metadata is entered for the IA, have those xml files parsed to add that information automatically to the flac tracks by some uber-script

Step 6 - Recompute the "file md5" signatures for the falc tracks that now have the id3 tags based on the IA metadata.

Step 7 - Add those flc-id3tagged-md5 signatures to the correct source in db.etree.org so the IA<-> db will link correctly

Step 8 - Submit to the Contribution ENgine so zip files etc are correct

The unknowns to me are:
1. Who has the script to take the metadata files created on the Internet Archive, parse them without errors and add the information as id3 tags to the correct flac file

2. Who can then make the new md5 signatures and get that added automatically to the correct shndi on db.etree.org

After doing the Phil collection BY HAND, to make this workable, we need really robust, automated scripts to use the metadata information entered/confirmed by the uploader

Step 5

Reply to this post
Reply [edit]

Poster: xtifr Date: Jun 17, 2005 5:52am
Forum: etree Subject: Re: How to automate FLAC file tagging

I thought db.etree used the flac fingerprint, not the whole-file md5? If so, there should be no problem, as changing the metadata doesn't affect the fingerprint. And if db.etree does use the whole-file md5, then it seems to me that this should be considered a bug in db.etree.

Also, once again, not a big deal, but flac doesn't use id3! Id3 is a specific tag format used by mp3 (and sometimes used with shorten). Flac uses its own tag format (actually, to be precise, it uses Ogg's tag format).

Reply to this post
Reply [edit]

Poster: Administrator, Curator, or StaffMatthew Vernon Date: Jun 17, 2005 6:48am
Forum: etree Subject: Re: How to automate FLAC file tagging

The Internet ARchive uses md5 signatures for most of its enternal workings as far as I know. It is NOT db.etree.org that requires it.

If there are multiple versions - one with and one without id3 tags, it would seem useful to have the md5 signature so you knew you were getting the one you want. The flac fingerprints would be the same and not distinguish versions that differed in their id3 tagging.

Also, in the past, but maybe not now, the md5 signatures were more useful than flac fingerprints for confirming that ALL of the intended files are present, not simply that the ones that are present pass the fingerprint check.

Reply to this post
Reply [edit]

Poster: xtifr Date: Jun 17, 2005 8:30am
Forum: etree Subject: not id3!

Again, NOT ID3! I realize I'm veering a little off-topic here, but the distinction is somewhat important, because you can attach id3 tags to flac files - in fact, you can attach id3 tags to any file. However, no software that I'm aware of will recognize id3 tags attached to flac files - just as no software will recognize (for example) id3 tags attached to a Microsoft Word document. Flac and Word have their own tag formats, which are not id3!

There are standalone programs that do id3 tagging (and can work with arbitrary files). The suggestion that flac uses id3 may lead to people using these programs to improperly tag their flac files with id3.

Reply to this post
Reply [edit]

Poster: Administrator, Curator, or StaffDiana Hamilton Date: Jun 17, 2005 7:27am
Forum: etree Subject: Re: How to automate FLAC file tagging

The db "uses" whatever data people have cared to put into the checksum fields for a given fileset. That can be whole-file md5s, flac fingerprints, wav md5s for the uncompressed wavs- sometimes all 3. It's tolerant enough to accept whatever people have on hand sufficient to distinguish the fileset they have.

Examples:
http://db.etree.org/shninfo_detail.php?shnid=29601
http://db.etree.org/shninfo_detail.php?shnid=21165

This post was modified by Diana Hamilton on 2005-06-17 14:27:52

Reply to this post
Reply [edit]

Poster: Spoogles Date: Jun 17, 2005 7:35am
Forum: etree Subject: Re: How to automate FLAC file tagging

foobar and tradersfriend plug in are a great way to tag all you flacs w/ metadata.

http://foobar2000.org

Reply to this post
Reply [edit]

Poster: xtifr Date: Jun 17, 2005 8:16am
Forum: etree Subject: Re: How to automate FLAC file tagging

> The db "uses" whatever data people have cared to put into the checksum fields for a given fileset.

Ah, ok, that makes more sense, thanks for the clarification.

Reply to this post
Reply [edit]

Poster: Administrator, Curator, or StaffDiana Hamilton Date: Jun 17, 2005 5:42am
Forum: etree Subject: Re: How to automate FLAC file tagging

2. Who can then make the new md5 signatures and get that added automatically to the correct shndi on db.etree.org

After doing the Phil collection BY HAND, to make this workable, we need really robust, automated scripts to use the metadata information entered/confirmed by the uploader


Good points all the way across! Relevant to these last ones, the db side additions would certainly have to be more automated than now. We have a "feature request" for that to happen for bt> db too, but I don't know when it will.

This post was modified by Diana Hamilton on 2005-06-17 12:42:27